Today almost unknown to non-lovers of old computers, this type of memory accounted for more than twenty years, from 1955 al 1975 circa, the most important (often the only) available technology for the manufacture of RAM from the acceptable reliability characteristics, size and cost per bit.
These memories were used in early computers before the spread of semiconductor memories.
The main technologies of the memories of computers in use in the 1970, shown on the cover of a special issue of the magazine Electronics World.
Early research on the memories to magnetic cores were performed by Au Wang and Way Dong Woo two researchers who developed the 1949 the pulse transfer controlling device. The name refers to a device that by means of the magnetic field should control the activation of an electromechanical system. Wang and Woo were working at the University of Havard, that unlike the MIT was not interested in promoting homegrown inventions. However Wang managed to get assign the invention patent.
Meanwhile, Jay Forrester at MIT was developing the Whirlwind computer and was aware of Wang's research and Woo. The Whirlwind computer required a very fast main memory (for the time) in order to run a flight simulator in real time. Forrester had initially thought of using Williams tubes for main memory, but these were inherently unreliable and difficult to manage.
OPERATION PIPE WILLIAMS
When a bright point is created on the screen of a cathode ray tube, the point remains on the screen for a certain period even after the signal that generated it has ceased, this is due to the presence of phosphors that induce an effect of luminous persistence. The Williams tubes utilize a different principle but that is reminiscent of the light persistence. Because the area of secondary emissions subject to irradiation of cathode ray changes polarity becoming positive and the immediately surrounding area becomes negative. The presence of negative or positive area could be detected by an external plate by electrostatic effect could detect the change in polarity when the cathode ray illuminated the point again. The reading was destructive because it erased the information stored. In addition, the polarity of the points with the decaying time and then were periodically rewritten.
Memory magnetic core
One difficulty was linked to the fact that access to the data caused the cancellation of the content read from the memory: Wang solved the problem by associating with each reading even writing in order to reinstall the stored data. While Forrester was able to use a small number of connections to handle a large number of magnetic cores.
The draft Forrester required that one of the magnetic cores wires crossed at an angle of 45 degrees, this could not be done automatically by machines and then the memories were assembled manually with the aid of a microscope and precision instruments.
At first the memories they cost about a dollar per item, but by the end they had come to cost about one US cent per element. These memories were still in the seventies replaced by semiconductor memories.
The magnetic core memories were part of a series of technologies related to the magnetic properties of materials. In the fifties thermionic valves were developed, these were sophisticated devices to be produced, fragile, relatively bulky and consumed a lot of power. The magnetic core memories, however, were small devices, solids and low consumption, They were often used for military applications seen their robustness. A famous example was the MOBIFIC computer developed by Sylvania for the American army.
Operation of the magnetic core memories
Detail of a magnetic core memory, the distance between the rings is 1 mm circa
The ferrite magnetic cores are arranged in a two-dimensional grid rings. The rings were paths from the wires that used to carry the memory management signals. In the first four-wire system is used; X,Y, Sense and Inhibit.
Each ring memorized a single digital state, each two-dimensional grid was accessible in a single clock cycle and allowed manipulation of a single ring. Stacking in an appropriate manner a number of grids you could get the reading or writing of a word in one clock cycle.
The magnetic rings based their ability to store information of the ferromagnetic hysteresis material, the rings subjected to a certain magnetic field tended to maintain a certain state magnetic until oncoming a new magnetic field of adequate intensity capable of reversing the magnetic field stored by material.
The switching from one state to another by sliding it had a sufficiently high current in the wires that run through the core.
A lower current does not, however, produces no change of state in the nuclei
The union of two semicorrenti was enough to change the magnetization state.
As mentioned, each ring was crossed by the X and Y lines: if you wanted to write a particular ring he was activated its line X and Y, just he saw that ring the entire magnetic field (sum than that brought by the X and Y lines) while the other rings were only seeing half of magnetic field or no magnetic field.
Therefore, only the addressed ring happened to be affected by a magnetic field sufficient to change the polarity of the magnetic field. The polarity of the field was defined by the polarity of the current that passed through the lines X and Y.
The current flowing in the wires induced an electromagnetic field that, when combined in the ring they altered the polarity.
Writing and reading
The reading in this kind of memory is a relatively complex. reading substantially consists in bringing 0 the addressed bit for reading. It piloting the X and Y lines at half power in the direction that causes the magnetic field that stores the 0 in memory. If the memory is actually 0 nothing is happening, but if the memory contained a magnetic field which indicated the bit 1 then on Sense line would read a short electrical pulse due to the change of polarity of the magnetic core.
Then the memory manager, After sending the appropriate line X and Y, He controlled the Sense line of the memory to verify the presence of 0 O 1. This read-destructive methodology was altered since the data always bringing the state memory 0.
Over the years he was added to a new thread that served to inhibit writing with a current opposite to that of magnetization nullifying a semicorrente.
The writing was following a similar principle, except that memory always is set to 1. The writing is assumed that the cells are already 0 due to a previous reading and activates the lines X and Y in order to produce a magnetic field variation that harbors the magnetic field of the magnetic core to the state 1.
If the cell is actually written 1 the procedure ends; if instead it should be placed in the line signal 0 It prevents writing through the line of inhibition. In this line it is made to slide a small current in the opposite direction that generates a magnetic field opposed to that generated by the X and Y lines and which prevents to change the state of the core thereby leaving the state memory 0.
Given that the Sense line and that of inhibition were not used simultaneously, it was decided to use a single line for both functions with an appropriate circuit that commuted the function of the line depending on whether there was a read or a write.
Since each reading also called for a writing, many computers took advantage of the fact. The computers included instructions able to pause the writing process after the one reading. This way if your computer had to read and then immediately write the same memory cell could do this in one step. The memory read data, it sent to the processor and placed on break. The processor and elaborating the data sent in new data that was written directly from memory during the write operation needed to regenerate the memory status. This made it possible to avoid unnecessary write speeds up your computer.
The COMPUTER CDC6600
Magnetic core memory of the CDC6600
The speed of these memories was of the order of Megahertz (speed similar to that of the computers of the eighties as the Commodore 64, The first systems had a read time of 6US, which it was later reduced to 1.2uS early seventies. At the end of the seventies he had come to have the models to 600ns.
The magnetic memories kept the state indefinitely without power. In addition, this type of memories is relatively little subject to electromagnetic interference and radiation. These are important characteristics in space and military applications fact, these memories were used for many years in these areas even after the spread of semiconductor memories. For example, the Space Shuttle computers initially used this type of memories. The magnetic memories survived the Space Shuttle Challenger disaster and allowed the Commission of Inquiry to have a report on what he measured the computer before the explosion.
A characteristic of magnetic memories is that the hysteresis loop is sensitive to temperature. Some companies like IBM kept their memories at an operating temperature above the ambient temperature with the heaters.
Memory card magnetic core from a Siemens computer (1964).
Given its historical significance and the events that gushed, Whirlwind (turbine) It deserves some additional record. At the end of a long and troubled history, Whirlwind proved highly innovative though still based on valve technology.
The project's origins date back to 1946 and they took its cue from the realization of a flight simulator, with analog technology, for training pilots of the Navy.
In addition to memory some of the most important inventions were nuclei:
monitor cathode ray with graphic display, operable with light pen;
massive use of the connected remote terminals on telephone lines and operating in real time;
techniques for increasing reliability and H24 for uninterrupted operation for every day of the year;
art software, between an operating system Trai first worthy of the name, and a programming language that somehow can be considered the father of FORTRAN
Hand in hand with the development of the Whirlwind, realized in a single copy, proceeded the project of SAGE (Semi Automatic Ground Environment) designed to defend the whole USA from possible air and missile attacks.
The final version of Whirlwind, It was in fact mounted AN / FSQ7 computer intended to equip SAGE in well 27 specimens. These 27 nodes were interconnected through a telephone network dedicated, they received and to prepare real-time signals of the search radar stations.
Beating stiff competition, IBM won the succulent contract for the construction of the whole project and is used SAGE, in addition, a substantial know-how luggage.
The implication began in 1952 e, in the maximum effort when more than 7000 People working on the SAGE project.
Among the technical staff it was also Kenneth Olsen who leave MIT in 1957 to found the DEC- Digital Equipemen Corporation and make it rise to the role of world lesder minicomputer.
In fact Digital ( term commonly used to DEC) He invented the minicomputer architecture of opening up a new market segment. The first successes were bound to the prolific series of PDP (Programmed Data Processor), PDP-1 1960 the PDP-11 of the seventies.